This proposal describes an ancillary study to a recently initiated comparative effectiveness trial Surgical Ablation versus No Surgical Ablation for Patients with Persistent or Longstanding Persistent Atrial Fibrillation Undergoing Mitral Valve Surgery (NCT00903370), which is part of the NHLBI-sponsored Cardiothoracic Surgery Network (CTSN). Its aims are to perform a time-sensitive substudy of continuous heart rhythm monitoring and to develop and apply novel statistical tools to assess ablation effectiveness in groups of patients. As such, it will provide critical new knowledge to support a secondary objective of the parent trial: to inform rhythm monitoring and analysis strategies for future trials of rhythm control in atrial fibrillation (AF). Great enthusiasm for innovative tissue-destroying (ablation) procedures to treat AF has led to technology outstripping science. Mechanisms of AF (which afflicts >3 million Americans) remain poorly understood, methods to assess its cure are primitive, and guidelines for analyzing rhythm outcomes are flawed due to lack of appropriate statistical methodology. Why? After ablation procedures, AF may recur in episodes of varying duration. Yet current monitoring consists of snapshots of semiannual rhythm recording (Holter), or more frequent-but short-transtelephonic monitoring (TTM). Ideally, an implantable monitoring device could detect all these episodes over a patient's lifetime; however, even if this were possible, statistical methods to analyze this series of occurrences in groups of patients for comparative effectiveness of AF therapies or to identify risk factors are poorly developed. Right now a window of opportunity exists to make important strides in more thorough AF assessment and proper data analysis because the CTSN has just launched a randomized trial of surgical AF therapies that includes weekly TTM of atrial rhythm as well as traditional 6- and 12-month Holter assessment. We propose a time-sensitive substudy of implanting an FDA-approved miniature loop recorder to capture AF episodes continuously. These data will be used in the development of a suite of new analytic methods for aggregating data such as these across time and across patients to characterize time course and identify modulating factors. They will then be used to assess the trial's rhythm endpoints and suggest designs for more informative future trials with substantially reduced sample size. We have assembled a team of clinicians and statisticians, including individuals who designed the parent trial, who are experienced in implanting the monitoring device, and who are experts in analytic methods research, to develop and apply novel methods to these several types of rhythm data. We will leverage developments from an intramural NHLBI machine learning program at Cleveland Clinic and State of Ohio-funded Atrial Fibrillation Innovation Center. The analytic methods have wide applicability to the important field of longitudinal data analysis.